Thinking in terms of expansion and split, characterised by the family tree model, comparative linguists typically regard a proto-language as if it were the result of rapid splitting during a period of punctuation. But it is probably not. A period of punctuation typically interrupts a long interval of equilibrium and the beginning of a language family is likely to lie in the type of linguistic situation that is produced by convergence within a linguistic area. Indeed, a language family may not have its basis in just one language.

§4.2 had preliminary discussion of how the putative ancestor language of a family should be seen as similar to the languages that are around us in the world today – with suppletions, irregularities, substrata, etc. Having now introduced the punctuated equilibrium model, we can take this discussion one stage further.

During a period of equilibrium contiguous languages diffuse features – phonological structures and systems (together with their phonetic realisations), grammatical categories, lexemes and (at a slower rate) grammatical forms. In a given area we may have two or three languages that have converged together in structure and, to an appreciable extent, in forms. They will belong to a number of distinct – although closely allied – political groups. They will be separate languages in terms of the strict criterion of intelligibility, although many members of the language community are likely to be bilingual or trilingual.

Suppose that the political and linguistic equilibrium in such an area is punctuated by the arising of a charismatic military or religious leader, or by the development of agriculture, or for some other reason.